University of Alberta, Applied Mathematics Institute, Department of Mathematical and Statistical Sciences
Canadian Applied Mathematics Quarterly
It is vital for natural gas Local Distribution Companies (LDCs) to forecast their customers' natural gas demand accurately. A significant error on a single very cold day can cost the customers of the LDC millions of dollars. This paper looks at the financial implication of forecasting natural gas, the nature of natural gas forecasting, the factors that impact natural gas consumption, and describes a survey of mathematical techniques and practices used to model natural gas demand. Many of the techniques used in this paper currently are implemented in a software GasDayTM, which is currently used by 24 LDCs throughout the United States, forecasting about 20% of the total U.S. residential, commercial, and industrial consumption. Results of GasDay'sTM forecasting performance also is presented.